I’d like to begin sharing my lessons learned with the community–a “practical” approach that I have been developing while working in the trenches of the Biotechnology industry. As a researcher-for-life, I observed what worked and what did not. After extracting the best practices that actually worked and taking out those that failed, I have been testing and refining the approach with my fellow scientists and engineers to continuously improve it. I’m a tinkerer so this will always be in a Beta version.
Of many, this is the first one. I’m calling it:
Lean & Agile Design of Experiments (DOE) for Quality by Design
1. The goal of the course is to map the “Design Space” in an effective & efficient way.
Most QbD submissions failed due to this — not correctly mapping a “Design Space.”
Design Space is the core element of Quality-by-Design approach. Properly doing this will provide financial benefits to the company–quality improvement, regulatory relief, faster time-to-market, etc.
2. Another important Aha! lesson for me: Knowledge is insufficient to transform behavior and practices. A business process must accompany the transformation to enable the knowledge. In R&D it was how R&D projects were managed. We need to change how R&D projects are run. (More on this soon). This course will provide the right project management fit for the ever-changing R&D.
Here are unique points of this course:
Practical – Agile
Designed with the scientist’s needs and R&D organization’s constraints in mind, Lean & Agile DOE for QbD provides tools you can use right away. Scientists will be able to handle constantly-changing project requirements with ease and finesse.
No Fluff – Lean
Scientists are always pressed for time. No philosophical discussions, pure practical knowledge and tools.
Shotgun approach to training is a waste. Only download the knowledge you need to meet your immediate project needs. Each class is compressed into 3-hour bite size.
Best way to learn is by doing. Each module has carefully designed exercises to generate Aha moments and transform knowledge into skills and insights.
Based on continuous research and experience, the refined approach delivers results every time.
Here is what each module covers:
Experiments are costly. They require investments of time, money and collaboration among different stakeholders. Thankfully there is an efficient way.
Experiment Canvas TM is a methodology which knits the essential elements of experimental planning and data analysis. A one-page map will guide scientists plan their experiments methodically but with ease. As an icing on the cake, one can communicate their experimental plan, progress, data, analysis and insights in less than 10 minutes.
PH (Process-Hypothesis) Map
When planning experiments, scientists must identify the critical parameters or X’s that determine critical quality attributes or Y’s. You and Subject matter experts will generate process maps that generate both a shared understanding of the process and a list of candidate hypotheses. This reference provides the basis for the PH map. Solving problems is an iterative and dynamic process that generates new knowledge. You will learn how to use PH Maps to capture current knowledge, create hypotheses and prioritize your next actions.
Dancing with Variation TM
Not all variations are the same. Understand why variation is important in any experiment. You will learn how to investigate and identify multiple sources of variation using graphical tools such as variability and controls charts. Then develop the correct control strategy based on the type of variation. This approach is highly effective for Root Cause Analysis, when searching for the sources of variation.
Probing before DOE
Before committing to a costly DOE, a scientist must assess whether the process is stable for a DOE. Otherwise, noise will trump signal in the data. To do this, hypotheses from the PH map need to be tested. Multiple hypotheses can be tested with a sampling plan, resulting in a more efficient process to find the answer to a problem. You will learn to translate the theories from your PH Map into a Sampling plan. After this you may not even need a DOE!
DOE Deep Dive
This module introduces the philosophy and theory behind the field of Design of Experiments (DOE). A hands-on case study demonstrates the benefits of a DOE approach over the common One-Factor-At a-Time (OFAT) experiment. DOEs use Factorial designs to gain the most knowledge from experiments. You will learn how to design and analyze Full and Fractional factorial designs. The module will then focus on the benefits of using Fractional factorials and the guidelines for selecting Fractional factorial designs. We will use a visual planning tool—Factor Tree—to communicate the data structure of your experiment.
DOE Design Space
QbD is about “Design Space.” This module delves into deeper topics of DOE such as center points, curvature, and the “Design Space.”
Signal or Noise? Measurement Check
How can you trust your data? Are you confident that the noise doesn’t trump the signal? Don’t make the mistake of misinterpreting your data. Learn the practical way of assessing and improving your measurement system.
Experiments help scientists make predictions. Accordingly, experiments need to be run under a range of conditions representative of future use. In fact, manufacturing processes are exposed to various noise factors. Noise factors are sources of variation, and this can significantly impact the results and conclusions of experiments. In this module, you will learn multiple strategies (i.e. blocking, replicates, repeats, randomization, etc.) which will help you manage noise and gain the most knowledge from your experiments.
This module takes Managing Noise one step further to leveraging noise and putting theory into practice. Other noise topics such as dealing with outliers are covered.
Lean and Agile ExperimentsTM
With many unknowns in a new product development, this module proposes a different way of running R&D projects—in a lean and agile fashion—minimizing bureaucracy.
If you’d like to request this course for your team, please email or leave a comment.